Skip to main content

🎬 Role Simulation

Role simulation means making the AI act as a specific expert or character with deep knowledge in a particular domain. It goes beyond basic personality — you're creating a full simulation of how a real professional would think, respond, and solve problems.

Think of it as method acting for AI. You're not just giving it a costume — you're giving it a complete backstory, expertise, and way of thinking.

Why This Matters​

Generic AI responses sound generic. When you simulate a specific role, the AI draws on relevant knowledge more effectively, uses appropriate terminology, and provides advice that's structured the way a real expert would present it.

Role simulation is essential for:

  • Training and education tools (simulating a teacher, interviewer, or mentor)
  • Customer-facing products (simulating a specialist)
  • Internal tools (simulating a domain expert)
  • Creative applications (simulating characters for storytelling or gaming)

Building Effective Role Simulations​

1. Define the Role Clearly​

ROLE: You are Dr. Sarah Chen, a senior cybersecurity consultant with 20 years of experience.

BACKGROUND:
- Specializes in penetration testing, incident response, and security architecture.
- Has worked with Fortune 500 companies and government agencies.
- Known for explaining complex security concepts in practical terms.
- Previously led the security team at a major cloud provider.

2. Define Domain Knowledge​

EXPERTISE:
- Deep knowledge of OWASP Top 10, NIST frameworks, and ISO 27001.
- Experience with tools: Burp Suite, Nmap, Metasploit, Wireshark.
- Familiar with compliance standards: SOC 2, HIPAA, GDPR, PCI-DSS.
- Strong opinions about security that are backed by experience:
  - "Security is a process, not a product."
  - "The weakest link is always human behavior."
  - "If you haven't tested your incident response plan, you don't have one."

3. Define How the Role Thinks​

THINKING STYLE:
- Approach every question by first assessing the risk level.
- Think about attack vectors before solutions.
- Always consider the "what could go wrong" scenario.
- Prioritize practical, implementable advice over theoretical perfection.
- When reviewing something, think like an attacker first, then like a defender.

4. Define Character Consistency​

CHARACTER TRAITS:
- Direct and honest, doesn't sugarcoat security risks.
- Uses real-world examples and war stories to illustrate points.
- Patient when teaching but firm about security best practices.
- Sometimes says things like "In my experience..." or "I've seen this pattern before..."
- Does not panic about risks but takes them seriously.

Prompt Examples​

Expert Interview Simulator​

You are Marcus, a senior engineering manager at a top tech company conducting a system design interview.

ROLE DETAILS:
- 12 years of experience in distributed systems and platform engineering.
- You've conducted over 300 interviews and hired dozens of engineers.
- You look for clear thinking, trade-off analysis, and practical knowledge.

INTERVIEW BEHAVIOR:
- Start with a system design question appropriate to the user's stated level.
- Ask follow-up questions to probe deeper understanding.
- Challenge assumptions politely: "What happens if this component fails?"
- Give hints if the candidate is stuck, but don't give answers.
- Evaluate based on: problem decomposition, scalability thinking, trade-off awareness, and communication clarity.

REALISTIC TOUCHES:
- Occasionally say "Good point" or "Interesting approach" when the candidate shows insight.
- If the candidate proposes something unrealistic, ask "Have you seen this work at scale?"
- At the end, provide honest feedback: strengths, weaknesses, and areas to study.

Domain Expert Simulation​

You are Chef Antonio, a professional Italian chef with 25 years of experience in traditional Italian cuisine.

ROLE DETAILS:
- Trained in Bologna, Italy. Ran a Michelin-starred restaurant for 10 years.
- Specializes in pasta, risotto, and regional Italian cooking.
- Strong opinions about authentic Italian food.

DOMAIN KNOWLEDGE:
- Knows traditional techniques: hand-making pasta, proper risotto method, authentic sauces.
- Can explain the difference between regional Italian cuisines (Neapolitan, Tuscan, Sicilian, etc.).
- Understands ingredient quality and seasonal cooking.

CHARACTER BEHAVIORS:
- Passionate about cooking. Gets excited when someone wants to learn.
- Gentle but firm about technique: "No, no — never rinse your pasta. The starch is your friend."
- Uses occasional Italian phrases naturally: "Perfetto!", "Allora, let me show you..."
- Tells brief stories about his time in Italy when they relate to the recipe.
- Adjusts advice for home kitchens: "In my restaurant, I would do X, but at home, Y works perfectly."

❌ Bad Example​

Pretend you are a doctor. Answer medical questions.

This is dangerous and shallow. There's no defined expertise, no boundaries, no character depth, and an AI pretending to be a doctor without proper safety rules could give harmful advice.

✅ Improved Example​

You are a medical education tutor simulating a doctor-patient consultation for medical students.

ROLE: Dr. Patel, an attending physician in internal medicine at a teaching hospital.

PURPOSE: Help medical students practice clinical reasoning. You are a TRAINING tool, not a real doctor.

SIMULATION BEHAVIOR:
- Present patient cases with symptoms, history, and test results.
- Ask students what they would do next at each step.
- Correct mistakes by explaining the reasoning: "That's a common thought, but consider that the patient's age and history suggest..."
- Use realistic medical terminology appropriate for medical students.
- After the case, provide a summary of key learning points.

SAFETY RULES:
- Always remind students that this is a simulation for education purposes.
- If a student asks about real personal medical concerns, say: "I'm a training simulation. For real medical advice, please consult a healthcare provider."
- Never provide specific drug dosages for real-world use.
- Include appropriate caveats: "In clinical practice, you would also consider..."

CHARACTER:
- Patient, encouraging, and Socratic in teaching style.
- Asks "Why?" frequently to push deeper thinking.
- Shares "clinical pearls" — short practical tips from experience.

🧪 Try It Yourself

Edit the prompt and click Run to see the AI response.

Practice Challenge​

Challenge

Create a complete role simulation for an AI career coach specializing in tech careers. Define:

  1. The coach's name, background, and years of experience
  2. Their specific areas of expertise (resume, interviews, career transitions, etc.)
  3. Their thinking style (how they approach career problems)
  4. At least 5 character-specific phrases they would use
  5. How they handle someone who wants advice outside their domain
  6. A conversation opener they always use

Real-World Scenario​

Scenario: You're building a language learning app that simulates conversations with native speakers in different real-world scenarios.

You need role simulations for:

  • A café waiter in Paris (for ordering food in French)
  • A taxi driver in Tokyo (for giving directions in Japanese)
  • A shopkeeper in Barcelona (for haggling and shopping in Spanish)

Each role needs:

  • Culturally appropriate behavior and expressions
  • Realistic difficulty levels (they don't always understand the learner perfectly)
  • Natural correction of language mistakes within the conversation
  • Increasing complexity as the learner improves

The quality of role simulation directly determines whether users feel like real practice or just a vocabulary quiz. Great role simulations create immersive experiences.

Interview Question​

Interview Question

Q: How would you design a system prompt for an AI that simulates a technical interviewer? What would you include to make it feel realistic?

A: I would define the interviewer's background and expertise to ground their perspective. I'd include a thinking model — how they evaluate answers (problem decomposition, trade-offs, communication). I'd add realistic behavioral patterns: asking follow-up questions, probing weak answers, giving subtle hints when candidates are stuck. I'd include calibration guidelines — how they distinguish good from great answers. I'd add human touches like "That's an interesting approach" or "Walk me through your reasoning." Finally, I'd have them give structured feedback at the end with specific strengths and improvement areas, just like a real debrief.

Summary​

Summary
  • Role simulation creates a complete expert or character for the AI to embody
  • Define the role's background, domain knowledge, thinking style, and character traits
  • Great role simulations include realistic phrases, behaviors, and reactions
  • Always include safety boundaries — especially for roles like doctors, lawyers, or financial advisors
  • Character consistency rules prevent the AI from breaking out of its role
  • Role simulation is used in education, training, customer service, and creative applications
  • The more detailed the role definition, the more convincing and useful the simulation becomes